French voice cloning is not as good as English

[baptx]

You can see that French voice cloning is not as good as English, for example by cloning the French president Macron voice by playing a part of this video and recording the speakers sound with Audacity to make him say a sentence in French (we don't recognize the voice of the person): https://www.youtube.com/watch?v=UTUXYricb1o And then clone the voice with a video where he speaks English to make him say a sentence in English (this time we recognize the voice of the person): https://www.youtube.com/watch?v=tc3yEJ3yBv8

I used the latest version with OpenVoice V2 (https://github.com/myshell-ai/OpenVoice/blob/main/docs/USAGE.md#openvoice-v2) on Linux and an Intel CPU / GPU. There does not seem to be a difference with WAV or OGG input file format.

Is there something to change to make it work better or can it be improved in a new version?